HAL Id: hal-02960344https://hal.archives-ouvertes.fr/hal-02960344

Submitted on 8 Oct 2020

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Experimental evaluation and modeling of thehydrodynamics in structured packing operated with

viscous waste oilsMargaux Lhuissier, Annabelle Couvert, A. Kane, Abdeltif Amrane, Jean-Luc

Audic, Pierre-Francois Biard

To cite this version:Margaux Lhuissier, Annabelle Couvert, A. Kane, Abdeltif Amrane, Jean-Luc Audic, et al.. Ex-perimental evaluation and modeling of the hydrodynamics in structured packing operated withviscous waste oils. Chemical Engineering Research and Design, Elsevier, 2020, 162, pp.273-283.�10.1016/j.cherd.2020.07.031�. �hal-02960344�

1

Experimental evaluation and modeling of the

hydrodynamics in structured packing operated with

viscous waste oils

Margaux Lhuissiera, Annabelle Couverta, Abdoulaye Kaneb, Abdeltif Amranea, Jean-Luc Audica, Pierre-

François Biarda

aUniv Rennes, Ecole Nationale Supérieure de Chimie de Rennes, CNRS, ISCR - UMR 6226, F-35000

Rennes, France

bUniLaSalle-Ecole des Métiers de l’Environnement, Campus de Ker Lann, 35170 Rennes, France

Graphical abstract:

Highlights:

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

1.3

1.4

0.5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

UG

(m s

-1)

L/G (kg kg-1)

Transformer oil flooding

0,8fl

0,6ufl

Loading zone (transformer oil)0.8×UG,Fl

0.6×UG,Fl

Air inlet

Air outlet

Structured Flexipac® packing

Experimental data for varying UG and UL

• Pressure drop• Loading points and flooding points

Modeling of the exp. data• Billet-Schultes

correlations

Predictive simulations using the Billet-Schultes correlations

• Loading and flooding points• Liquid holdup hL• Interfacial area a°• Pressure drop DP

Scale-up of an industrial column

• 4000 Nm3 h-1

• 1 ≤ L/G ≤ 5 kg kg-1

Viscous solventsPDMS 20 (20 mPa s)Transformer oil (19 mPa s)Lubricant (79 mPa s)

1 m

120 mmID

DP

Step 1

Step 2

Step 3

Step 4

Example

Jour

nal P

re-p

roof

2

Hydrodynamics of two waste oils and PDMS in a structured packing was investigated

Pressure drop in the loading zone was lower than 450 Pa m-1

Loading and flooding velocities were relatively low, especially using the viscous lubricant

Billet-Schultes correlations were efficient to predict the loading and flooding points and the

pressure drop

Scale-up calculations proved that transformer oil can be used in an industrial scale packed

column

Abstract :

The purpose of this work was to study the hydrodynamic behavior of two viscous waste oils (a

transformer oil and a lubricant characterized by viscosities of 19 mPa s and 79 mPa s, respectively)

and a silicone oil (20 mPa s) in a laboratory-scale packed column (Dcol = 0.12 m). The column was

filled with structured packing made of corrugated sheets (Flexipac® 500Z HC) and was operated at

counter-current. Thus, the gas superficial velocities at the loading point were in the range from 0.40 to

0.65 m s-1 for liquid loads between 1 and 24 m3 m-2 h-1, and, at the flooding point from 0.56 to 1.07 m

s-1 for liquid loads between 6 and 36 m3 m-2 h-1. Both loading and flooding points were particularly

influenced by the solvent viscosity, leading to a narrow loading zone for the most viscous solvent

(lubricant). The pressure drop values remained reasonable, lower than 450 Pa m-1 in the loading zone,

even for the lubricant. Billet-Schultes correlations were used for the prediction of the loading and

flooding velocities and of the pressure drop. The specific constants of the model were determined.

These correlations enable accurate predictions of the loading and flooding points, with an average

relative error around 7-8%, and of the pressure drop in the loading zone, with an average relative error

of 15%. Simulations were performed with the Billet-Schultes correlations and showed that high liquid

holdup and interfacial area would be obtained with these viscous solvents in the selected packing.

Scale-up calculations proved that it would be possible to implement the transformer oil at industrial

scale in a packed column filled with the studied structured packing.

Keywords: Absorption; Hydrodynamics; waste oils; PDMS; packed column; modeling

Jour

nal P

re-p

roof

3

Nomenclature

a: specific surface area of packing (m2 m-3)

ah : hydraulic surface area (Eq. 6 and Eq. 7, m2 m-3)

a°: interfacial area relative to the packing volume (m2 m-3)

ARE: Average Relative Error

CP, CLo, CFl, Ch : constants relative to each commercial packing according to Billet-Schultes

dh: hydraulic diameter = 4/a (m)

Dcol : column diameter (m)

F: volume flowrate (m3 h-1 or m3 s-1)

g: specific gravity constant (9.81 m s-2)

G: gas mass flowrate (kg s-1)

hL: liquid holdup (-)

L: liquid mass flowrate (kg s-1)

Re: Reynolds number (-)

U: superficial velocity (m s-1)

Z: packing height (m)

RE: Relative error

Greek letters

ΔP/Z : Linear pressure drop (Pa m-1)

: Packing void fraction (-)

Jour

nal P

re-p

roof

4

: resistance coefficient (-)

dynamic viscosity (Pa s)

density (kg.m-3)

Subscripts

L: Relative to the liquid phase

G: Relative to the gas phase

Lo: At the loading point

Fl: At the flooding point

1. Introduction

Volatile Organic Compounds (VOCs) have harmful effects on the environment through their global

warming contribution. Disturbing the Chapman cycle in the atmosphere by reacting with radicals,

VOCs cause tropospheric ozone accumulation and thus intensify global warming (Le Cloirec, 2004).

Besides, some VOCs could be toxic towards human health. Their impact on liver, blood and nervous

systems have been demonstrated (Kampa and Castanas, 2008).

Several mature technologies can be implemented to remove VOCs from air such as adsorption on

activated carbon filters, thermal and catalytic oxidations, absorption (gas-liquid scrubbing) or

biofiltration (Ruddy and Carroll, 1993). To target hydrophobic VOCs, absorption in a non-aqueous

phase (NAP) using silicone oils, phtalates, adipates or even ionic liquids, would be a promising

treatment (Hadjoudj et al., 2004; Heymes et al., 2006; Bourgois et al., 2006, 2009; Darracq et al.,

2010a; Darracq et al., 2010b; Guihéneuf et al., 2014; Biard et al., 2016, 2018; Rodriguez Castillo et

al., 2018, 2019). Nonetheless, all these commercial or laboratory made solvents are very expensive,

which would increase the CAPEX of an industrial process. Consequently, several studies have

Jour

nal P

re-p

roof

5

recently investigated the potential of cheaper NAP, such as waste oils (Bay et al., 2006; Lalanne et al.,

2008; Ozturk and Yilmaz, 2006).

In a previous paper investigating the physico-chemical properties (viscosity, volatility, partition

coefficient for toluene and dichloromethane) of several industrial waste oils (Lhuissier et al., 2018),

lubricant and transformer oil were proved to be potential NAP candidates for VOC absorption.

Nevertheless, these waste oils are from 10 to 70 times more viscous than water and have a low

surface tension, which can affect significantly their hydrodynamics, especially in irrigated packed

columns which are the most commonly used contactors at industrial scale for gas scrubbing (Le

Cloirec, 2004; Minne, 2017).

In a counter-current packed column, it is recommended to operate in the loading zone located

between the loading and flooding points (Billet and Schultes, 1999). Indeed, for gas velocities below

the loading points (in the pre-loading zone), the downward flow of liquid, and consequently the liquid

holdup, depend only on the liquid rate and are not influenced by the gas velocity, leading to low

interfacial area and gas-liquid interactions (Billet and Schultes, 1999, 1995, 1993; Heymes et al.,

2006; Mackowiak, 2010; Stichlmair et al., 1989). In this pre-loading zone, the pressure drop increases

linearly with the gas superficial velocity (Piché et al., 2001a). In the loading zone, the shear forces in

the gas flow support some of the descending liquid allowing to increase advantageously the liquid

holdup and the interfacial area (Billet and Schultes, 1999; Piché et al., 2001a). In the loading zone, the

pressure drop increases with the gas superficial velocity to a power of around 1.8 (Piché et al., 2001a).

Then, when the flooding point is reached, the shear stress of the gas flow is high enough to maintain

the liquid at the top of the column, i.e. the liquid starts to overflow and the pressure drop increases

sharply (Billet and Schultes, 1999, 1995, 1993; Piché et al., 2001b). Consequently, for a given gas

flowrate, the choice of a working point in the loading zone allows subsequently to determine the

diameter of the contactor (Roustan, 2003). More precisely, a working gas superficial velocity equal to

70-80% of the flooding velocity is recommended by several authors (Billet and Schultes, 1999;

Maćkowiak, 1991), even if a lower boundary of 60% was also proposed (Roustan, 2003).

Several semi-empirical correlations have been developed in the literature to determine the loading and

flooding points as-well-as liquid holdup and pressure drop in the pre-loading and loading zones of

either random and structured packing (Maćkowiak, 1990, 1991; Miyahara et al., 1992; Rocha et al.,

Jour

nal P

re-p

roof

6

1993; Billet and Schultes, 1999; S. Piché et al., 2001a, 2001b, 2001c; Heymes et al., 2006). These

correlations take into account the column diameter, packing characteristics (porosity, particle diameter,

specific surface area, etc.), liquid and gas properties (viscosity, density, superficial tension) and liquid-

to-gas mass flowrate ratio L/G. Besides, the pressure drop in the loading zone depends also on the

selected liquid and gas velocities. Most of these correlations were established using data gathered

with water-like solvents having low or moderate viscosities. The Billet-Schultes correlations were

developed in the 90ties for both random and structured packing with solvents having dynamic velocities

up to around 100 mPa s (Billet and Schultes, 1999, 1993). In 2006, Heymes et al. showed that these

correlations were the most accurate to predict the pressure drop of dumped packing using DEHA. In

2017, Minne also proved the good predictive capacity of these correlations using solvents with

different viscosities, densities and surface tension (ethylene glycol, silicone oil, water and Isopar G).

The procedure developed by Billet-Schultes involves the calculation of liquid holdup, which is the ratio

of the liquid within the column volume to the packing volume.

Only a few studies have been performed to assess the hydrodynamics of viscous solvents in packed

columns, and up to now, no one has been dedicated to waste oils (Brunazzi et al., 2002; Darracq,

2011; Guillerm et al., 2016; Heymes et al., 2006; Minne, 2017; Minne et al., 2018; Tsai et al., 2009).

Thus, the purpose of this work is to study the hydrodynamics of two waste oils, a lubricant and a

transformer oil, in a laboratory scale packed column, by comparison to another reference solvent: a

commercial silicone oil (PDMS 20). A modern 4th generation structured packing made of metallic

corrugated sheets was used (Flexipac® 500Z HC). Both the loading and flooding points for several

liquid-to-gas mass flowrate ratios (L/G) have been determined, as well as the pressure drop evolution

with the gas and liquid velocities. Besides, the reliability of the Billet-Schultes correlations was

evaluated allowing to subsequently simulate different conditions and to assess the scale-up of the

process at industrial scale.

2. Material and method

2.1 Chemical products

Jour

nal P

re-p

roof

7

The silicone oil used was composed of polydimethylsiloxane (PDMS), a synthetic oil named Rhodorsil

47V20 provided by Bluestar Silicones (Table 1). The waste oils were selected among those collected

by Chimirec (Javené, France).

2.2 Pilot-scale packed column

The irrigated packed column (120 mmID (Dcol) and packing height Z of 1 m) was operated at counter

current (Fig. 1). The air flow (G = 1.17 kg m-3 at 25°C and 1 atm) was introduced at the bottom of the

column and the liquid solvent was introduced at the top of the column. The column was filled with

structured packing (Flexipac®, provided by Koch Glitsch-USA) made out of corrugated sheets

arranged in a crisscrossing configuration to create flow channels for the gas phase (Table 2).

The air flowrate was regulated by means of a membrane valve located after a fan and measured by a

rotameter (GF Type SK 20 CH-8201, Switzerland). The column was fed with the liquid absorbent by

means of a centrifugal pump (Iwaki MD100, Japan). The liquid flowrates were regulated with a valve

and measured by a previously calibrated rotameter (GF type SK 11 CH-8201, Switzerland). The

experimental conditions and the range of operating parameters are shown in Table 3. For each liquid

flowrate selected, the gas flow was increased incrementally up to the flooding point (determined by

visual observation), and pressure drops (ΔP) were measured three times for each point using a

vertical U-shaped tube filled with water. Only the pressure drop corresponding to the packing was

measured. The temperature of the liquid phase ranged from 294 to 298 K, and the gas temperature

was kept constant at 298 K by means of a heat exchanger. Considering all the experiments, the

difference between the inlet temperatures of the gas and liquid phases was always lower than 3 K.

The accuracy and reliability of this set-up was previously checked by Darracq and Guillerm (Darracq,

2011; Guillerm, 2017).

Billet-Schultes correlations Billet and Schultes (B-S) developed a set of correlations allowing to determine (i) both the loading and

flooding gas superficial velocities (UG,Lo and UG,Fl) for a given L/G ratio and (ii) the pressure drop, liquid

holdup and the interfacial area for the selected working gas superficial velocity (UG) in both the pre-

loading and loading zones (Billet and Schultes, 1999, 1995, 1993, 1991). These correlations,

presented in Table 4, involve several constants which depend specifically on the packing (CP, CLo, CFl

and Ch). The structured packing used in this study was not previously tested by Billet-Schultes.

Jour

nal P

re-p

roof

8

Therefore, the constants used in our calculations were deduced from the experimental results (part

3.3). For a given packing, the loading and flooding points depend on the selected L/G ratio, the liquid

and gas properties and the packing properties. They can be deduced from the equations 1-7 and 8-11,

respectively. These sets of equations need to be solved by iteration (for example using the Excel®

Solver). The determination of the loading points involved the determination of the ratio of the hydraulic

to the geometric surface area (ah/a) according to Eqs. 6 or 7. This ratio must not be confused with the

ratio of the specific interface between the phases to the geometric area (a°/a), determined with Eqs.

18-20. Besides, these equations involve several resistance coefficients () whose calculation depends

on the nature of the dispersed phase (Eqs. 3-4 and Eqs. 10-11).

3. Results and Discussion

3.1 Experimental pressure drop of the irrigated column

The experimental pressure drop of the irrigated packed column increases logically with both the liquid

and gas superficial velocities according to Fig. 2. The gas velocity can be easily converted to the vapor

capacity factor FV, often used to report data dealing with the hydrodynamics of an irrigated packed

column (Minne, 2017), by multiplication with the square root of the air density (which was almost

constant in this study and equal to 1.17 kg m-3). Lower liquid velocities were applied for the lubricant

since the flooding was reached at a lower L/G ratio for a given gas velocity than for silicon and

transformer oils. This observation is consistent with the fact that both loading and flooding points

appear at lower gas velocities for viscous solvents at similar liquid densities (Minne, 2017).

Nonetheless, even with a high viscosity, it would be feasible to use this lubricant in a packed column.

In Fig. 2, experimental points located in both pre-loading and loading zones are represented. Some

values of the pressure drop were also measured in the flooding zone but are not represented in Fig. 2

for clarity. In the pre-loading zone, the pressure drop increases linearly with the gas velocity and is

minimally influenced by the nature and the flowrate of the liquid. Minne (2017) and Brunazzi et al.

(2002) also observed this behavior for various liquids with similar densities but very different

viscosities for random and structured packing, respectively. It is consistent with the fact that gas and

liquid have almost no interaction in the pre-loading zone. However, this behavior completely changes

Jour

nal P

re-p

roof

9

when the loading point is reached. Indeed, the pressure drop at given liquid and gas velocities is

significantly higher for the lubricant in the loading zone than for the transformer oil and the silicon oil.

Furthermore, the pressure drop increases with the gas velocity to the power of 1.8. All these

observations in the loading zone are in agreement with the literature (Minne, 2017; Minne et al., 2018;

Piché et al., 2001a). The results emphasize that the pressure drop values stay advantageously under

450 Pa.m-1 in both the pre-loading and loading zones for each of the viscous liquid phases, showing

that the implementation of viscous solvents would not affect significantly the operating costs (OPEX).

However, the results clearly show that lower L/G ratios must be selected when the solvent viscosity

increases to avoid loading and flooding points that would be too low, highlighting that the

determination of the loading and flooding points would be determinant when designing a column. To

reach this goal, robust models must be implemented.

3.2 Determination of the experimental loading and flooding

points

The log-log plots corresponding to Fig. 2 are provided as supplementary material (Fig S.1). For the

three solvents, the experimental loading and flooding superficial velocities, summarized in Table 5 and

Table 6, were determined from the break-up of the slopes between the pre-loading zone, the loading

zone and the flooding zone on these log-log plots.

The loading and flooding points for a given liquid-to-gas mass flowrate ratio (L/G) were close to the

results obtained previously by Guillerm et al. (2016) with the same packing using PDMS 50 (viscosity

of 50 mPa.s). The observed loading and flooding velocities are relatively low, in the range 0.39-0.65

and 0.56-1.06 m s-1, respectively. This behavior is justified by both the nature of the packing, which is

a structured packing with a high interfacial area leading to a high liquid capacity, and the high viscosity

of the solvents investigated.

Indeed, in agreement with the experimental observations and theoretical considerations of other

authors (Billet and Schultes, 1995; Brunazzi et al., 2002; Minne et al., 2018), both loading and flooding

velocities decreased with the solvent viscosity because of higher frictional forces. Thus, lower liquid

loads were operated for the most viscous solvent, i.e. the lubricant. Nonetheless, having similar

Jour

nal P

re-p

roof

10

viscosities, densities and surface tensions, the loading and flooding velocities for the transformer oil

and the silicon oil are not significantly different at a given liquid load. On the one hand, the influence of

the density on the hydrodynamics is rather well described in the literature. Indeed, the gravitational

forces are counterbalanced by the frictional forces at lower velocities when the density decreases

(Minne, 2017). On the other hand, the possible influence of the surface tension is still controversial in

the literature and seems to depend on the nature of the packing and of the liquid load (Minne, 2017).

Anyway, the Billet-Schultes model, applied in the next part, does not take the influence of the surface

tension into account.

3.3 Application of the Billet-Schultes: determination of the

specific constants for Flexipac® 500Z HC packing.

Eqs 8-11 show that the determination of the flooding point at a given L/G ratio depends on the

properties of the liquid phase (viscosity and density ) and on the properties and the nature of the

packing through its void fraction , specific area a and through a specific constant CFl. By trying to

minimize the sum of the relative errors (RE) between the experimental gas flooding velocities and

those deduced from the B-S correlations, the value of CFl was determined (Table 7). The value found

is in agreement with the range previously determined by Billet and Schultes (Billet and Schultes,

1999). The low average relative error (ARE) of 7.5% (calculated for 11 values) between the

experimental and theoretical gas flooding velocities (summarized on Table 6) shows that B-S

correlations are efficient for the prediction of the flooding points even for solvents with very different

properties. Minne (2017) found similar RE for a silicon oils using two dumped packing. The parity plot

is provided as supplementary material and does not put in evidence any systematic error (Fig. S-2(b)).

From Eqs 8-11, the liquid holdup at the flooding points was also determined (Table 6).

Eqs. 1-7 show that the determination of the loading point at a given L/G ratio depends on the

properties of the liquid phase (viscosity and density ) and on the properties and the nature of the

packing through its void fraction , specific area a and through two specific constant CLo and Ch.

Several couples (CLo ;Ch) allowed to minimize the ARE between the experimental gas loading

velocities and those deduced from the B-S correlations. Thus, the best couple summarized on Table 7

was determined through the simultaneous minimization of the ARE of both the experimental and

theoretical pressure drops and loading points. Only the pressure drops measured in the loading zone

Jour

nal P

re-p

roof

11

were considered. Indeed, the pressure drop in the pre-loading zone is low and suffers from higher

experimental uncertainties. Besides, Minne (2017) pointed out the relatively low performance of the

Billet-Schultes model to predict the pressure drop on the pre-loading zone.

The values of Ch and CLo found are in agreement with the ranges previously determined by Billet and

Schultes (Billet and Schultes, 1999), even if the value of CLo is slightly lower than the range of Table 7.

The low ARE of 8.0% (calculated for 10 values) between the experimental and theoretical gas loading

velocities (summarized on Table 5) shows that the B-S correlations are efficient for the prediction of

the loading points, even for solvents with very different properties. The parity plot is provided as

supplementary material and shows that the model slightly overestimates the loading points for the

lubricant and slightly underestimates them for the transformer oil (Fig. S-2(a)), but these discrepancies

remain acceptable.

The results summarized in Table 5 and Table 6 clearly confirm that the liquid holdup would be higher

at a given L/G ratio for the lubricant, i.e. for a higher viscosity, for both the loading and flooding points.

Thus, even if viscous solvents must be operated at a lower superficial gas velocity and/or a lower L/G

ratio (i.e. lower absorption rate (Biard et al., 2018)), a higher interfacial area would be expected.

Finally, the value of Cp (Table 7) was determined through the minimization of the ARE between the

experimental and theoretical (Eqs 12-17) pressure drops of the points located in the loading zones.

The value of Cp found is in agreement with the range previously determined by Billet and Schultes

(Billet and Schultes, 1999). The low average relative error of 15.1% (calculated for 134 values: 42 for

PDMS 20, 32 for transformer oil, 60 for lubricant) between the experimental and theoretical pressure

drops in the loading zone shows that B-S correlations estimate with a good confidence level the

pressure drop even with viscous solvents. On the pre-loading zone, the agreement is lower, with an

ARE of 34.1%, which is consistent with the observations of Minne (2018). Nonetheless, it is not

recommended to operate on the pre-loading zone (Billet and Schultes, 1999). The agreement between

the experimental and theoretical pressure drops on the pre-loading and loading zones can be

assessed Fig. 2. The detailed ARE for the three solvents in each zone are summarized in Table 8. It

shows that the best agreement is found for PDMS 20 and transformer oil. For the lubricant, the ARE is

higher, around 27%, but remains fully acceptable. Furthermore, the parity plot of the pressure drop is

provided as supplementary material and does not put in evidence any systematic error (Fig. S-3).

Jour

nal P

re-p

roof

12

3.4 Discussion and simulations

From the constants determined previously (Table 7) and the B-S correlations, it was possible to

estimate the loading and flooding points for the two waste oils for an increasing L/G ratio in the 0.5-15

range (Fig. 3).

The loading and flooding velocities obviously decrease with the L/G ratio. For the lubricant (i.e. the

most viscous solvent), the loading velocity is from 6% (at L/G = 0.5) to 20% higher (at L/G = 15) and

the flooding velocity is from 10% (at L/G = 0.5) to 30% higher (at L/G = 15) than for the transformer oil.

Because of almost identical properties, the data for PDMS 20 are similar to those of the transformer oil

and are not represented. For the sake of comparison, a simulation for water was also carried out (Fig.

S-4). The results show that for this solvent, the loading and flooding points would be respectively from

17 to 52% higher and from 25 to 48% higher than for the transformer oil.

Fig. 3 shows that the choice of a working velocity at 80% of the flooding velocity allows operating in

the loading zone in any case. However, the choice of a working velocity at 60% of the flooding

velocity, proposed as a lower boundary by some authors (Roustan, 2003), would be inappropriate,

especially at high L/G ratio and for the lubricant. The breakthrough of the curves observed at a L/G

ratio around eleven is due to a shift of the dispersed phase (from the liquid to the gas phase)

corresponding to a value of 𝐿

𝐺× √

𝜌𝐺

𝜌𝐿 higher than 0.4 (Table 4).

Fig. 3 also highlights that the selection of an excessive L/G ratio would be inappropriate for several

reasons. Indeed, for the viscous solvents investigated, it would impose to select a low superficial gas

velocity, i.e. to increase the column diameter for a given gas flowrate to treat. Moreover, a low

superficial gas velocity would affect the turbulences in the gas phase, decreasing the gas-phase mass

transfer rate. Since a significant part of the mass transfer resistance can be located in the gas phase

during hydrophobic VOC absorption, even using viscous solvents (Biard et al., 2018; Rodriguez

Castillo et al., 2019), would decrease the overall mass transfer rate. Finally, for a high L/G ratio, the

loading zone is narrower. Thus, low variations of the gas velocity may induce significant variations of

the hydrodynamics within the column, which would complicate the monitoring of a packed column at

industrial scale.

Jour

nal P

re-p

roof

13

The evolution of the pressure drop DP/Z (orange curves), the liquid holdup hL (yellow curves) and the

ratio a°/a was simulated for several liquid loads (Fig. 4). hL and the ratio a°/a remain constant in the

pre-loading zone in agreement with the experimental observations of several authors with structured

packing (Billet and Schultes, 1993; Brunazzi et al., 2002; Zakeri et al., 2012). In the pre-loading zone,

hL and the ratio a°/a increase by 100% for the transformer oil when the liquid load increases by a

factor of 4 (from 5 to 20 m h-1). Moreover, hL and the ratio a°/a are around 30% higher for the lubricant

in the pre-loading zone. Brunazzi et al. (2002) also showed that the liquid holdup increases with the

solvent viscosity even in the pre-loading zone. Thus, a viscous solvent exhibits advantageously a

higher interfacial area although the pressure drop is almost insensitive to the nature of the solvent, at

least in the pre-loading zone.

When the loading zone is reached (around 0.60 m s-1 for UL = 5 m h-1 and around 0.45 m s-1 for UL =

20 m h-1 for the transformer oil; around 0.55 m s-1 for UL = 5 m h-1 for the lubricant), the gas and the

liquid start to interact. Thus, both hL and the ratio a°/a increase until the flooding point is reached

(around 1.20 m s-1 for UL = 5 m h-1 and around 0.78 m s-1 for UL = 20 m h-1 for transformer oil; around

1.02 m s-1 for UL = 5 m h-1 for the lubricant). A rather high liquid holdup, from 25 to more than 45%,

would be reached at the flooding point. High liquid hold ups are concomitant with a good wetting of the

column and with high interfacial area a°. The interfacial area can be even higher that the geometrical

area a of the packing according to the experimental observations of Tsai et al. (2009) on similar

structured packing fed with solvents having viscosities up to 14 mPa s. Cherif et al. (2017) showed the

good predictive capacity for a° of the Billet-Schultes correlations for a structured packing. In the

loading zone, contrarily to the pre-loading zone, the pressure drop is particularly sensitive to the liquid

load and the viscosity of the solvent. Thus, for example, at a liquid load of 5 m h-1 and a gas velocity of

0.8 m s-1, the interfacial area and the pressure drop for the lubricant would be around 40% and 15%

higher than for the transformer oil, respectively.

3.5 Packing scale-up

Jour

nal P

re-p

roof

14

According to part 3.3, the Billet-Schultes correlations provide fair estimations of the pressure drop and

loading and flooding points. These parameters are crucial for the evaluation of the operating cost

through the determination of the packed column diameter and the selection of the L/G ratio (part 3.4).

In this part, in order to study the scale-up of a column with this specific packing, a realistic case

involving a flowrate of 4000 Nm3.h-1 of air to be treated was considered (where N stands for the

standard temperature and pressure (STP) conditions, meaning 1 bar and 0°C according to the

IUPAC). The transformer oil was selected among the three solvents considering the observations of

the parts 3.3 and 3.4. Biard et al. (Biard et al., 2018; Rodriguez Castillo et al., 2019) already studied

the hydrodynamics of DEHA (12.5 mPa s), PDMS (50 mPa s) and two ionic liquids (50-60 mPa s) in a

packed column (Dcol = 1 m) filled with bulk Pall® rings for with the same gas flowrate using the Billet-

Schultes correlations. Apart from a different packing nature (corrugated sheets vs. random rings),

Pall® rings offer a geometric surface area of only 139.4 m2 m-3, around 3 times lower than

Flexipac®500 Z HC. Thus, both the computed loading and flooding superficial velocities for similar L/G

ratio (in the range 1-5) are significantly lower with the Flexipac® packing. Thus, it was impossible to

select the same working velocity of 1.52 m s-1 as Biard et al. (Biard et al., 2018; Rodriguez Castillo et

al., 2019) since this one is higher than the computed flooding velocity for the Flexipac® packing

whatever the L/G ratio considered.

Thus, for this scale-up assessment, a L/G ratio in the range 1-5 to avoid flooding velocities that were

too low (Fig. 3 (a)) was selected. Then, a working gas velocity taken at 80% of the flooding velocity as

recommended in the literature was considered (Billet and Schultes, 1999). A working gas velocity

(which decreases with the L/G ratio), in the range 0.54-0.86 m s-1, and a column diameter (which

increases with the L/G ratio), in the range 1.33-1.68 m, were calculated. Fig. 5 presents the evolution

of the liquid holdup hL, interfacial area a° and pressure drop DP/Z calculated with the B-S correlations

for increasing L/G ratios.

The calculated pressure drops would be in a reasonable range (95-175 Pa m-1) for an industrial

application. The pressure drop would be lower at industrial scale at given liquid and gas velocities than

at the lab-scale, because of negligible wall effects. These wall effects are taken into account by the

Billet-Schultes correlations through the column diameter (Eqs. 15-17). The pressure drop decreases

with the L/G ratio and with the working gas velocity UG, even if the liquid holdup and the liquid load

Jour

nal P

re-p

roof

15

increase. The low values of the pressure drop must be balanced with the low gas velocity tolerated by

this packing. It confirms that viscous solvents do not necessarily lead to high pressure drop (Biard et

al., 2018). Advantageously, the computed interfacial area is significant, higher than 200 m2 m-3. These

results emphasize the good hydrodynamic performance offered by this packing. Nonetheless, the use

of a low gas velocity would be detrimental for the mass transfer rate in the gas phase, in which the

main part of the resistance can be located, especially during the absorption of hydrophobic VOCs for

which the selection of viscous absorbents is justified (Rodriguez Castillo et al., 2019).

4. Conclusion

The purpose of this work was to study the hydrodynamic behavior of two viscous waste oils (a

transformer oil and a lubricant characterized by viscosities of 19 mPa s and 79 mPa s, respectively)

and a silicone oil (20 mPa s) in a laboratory-scale packed column (Dcol = 0.12 m) filled with structured

packing (Flexipac® 500Z HC). First, the results showed that it was possible to successfully apply such

viscous solvents in structured packing made of corrugated sheets. The loading and flooding points

were determined for L/G ratios between 0.3 and 14.0. Pressure drop in the loading zone values were

in the range from 50 to 450 Pa.m-1. The pressure drop values were significantly higher in the loading

zone at the same gas and liquid velocities for the most viscous solvent (lubricant) than for the other

solvents. Besides, lower liquid loads (i.e. liquid velocities) must be applied for the lubricant, since the

flooding point was reached at lower gas velocities for a given L/G ratio. Considering that the high

lubricant viscosity will also lower the mass transfer rate (directly, through lower turbulences, and

indirectly, through lower diffusion coefficients (Rodriguez Castillo et al., 2019)), the selection of such a

viscous solvent would be justified only for pollutant/solvent systems characterized by low partition

coefficient (i.e. high affinities).

In order to apply waste oils in a scrubbing process, the accurate prediction of the hydrodynamic

performance is a prerequisite. The Billet-Schultes correlations (Billet and Schultes, 1999) were used to

predict the loading points and flooding points, as well as the pressure drop, the liquid holdup and the

interfacial area at the working velocity. The specific constants of the packing used were determined by

numerical resolution. The ARE between the experimental and theoretical pressure drops in the loading

Jour

nal P

re-p

roof

16

zone was around 15%. Furthermore, the Billet-Schultes correlations successfully determined the

loading and flooding points with ARE around 7-8%. Several simulations were performed with these

correlations to assess the influence of the operating conditions. It highlighted the high hydrodynamic

performance of this packing. It also demonstrated that the use of viscous solvents does not

necessarily lead to excessive pressure drop in agreement with the experimental results. A fortiori, high

liquid holdup and interfacial area would be advantageously obtained. Nonetheless, low loading and

flooding velocities were computed, even for a low L/G ratio, which leads to select low working gas

velocities, leading to high diameter columns and lowered mass transfer rate in the gas phase. Finally,

the scale-up of a realistic packed column for gas treatment was assessed and showed that that it

would be possible to apply these kinds of solvents with the Flexipac®500Z HC packing even at

industrial scale. For a complete techno-economic assessment, the determination of the mass transfer

coefficients in both the liquid and gas phases will be necessary, allowing to evaluate concomitantly the

pressure drop, the overall mass transfer coefficient and thus, the resulting removal efficiency in a

given packed column (Biard et al., 2018; Rodriguez Castillo et al., 2019).

5. Acknowledgments

We are very grateful to the French Association Nationale de la Recherche et de la Technologie

(ANRT) for the CIFRE PhD grant N° 2016/0238 attributed to Margaux Lhuissier, and also to the

French governmental agency ADEME for the CORTEA funding n°1881C0001. We would like to thank

our industrial partner Chimirec for providing the waste oils used in this study. We would like to give a

warm thanks to Thomas HULL (UniLaSalle-Ecole des Métiers de l’Environnement) for proofreading

this article.

Declaration of interests

The authors declare that they have no known competing financial interests or personal relationships that

could have appeared to influence the work reported in this paper.

Jour

nal P

re-p

roof

17

References

Aluyor, E.O., Ori-Jesu, M., 2009. Biodegradation of mineral oils–A review. African Journal of

Biotechnology 8. Bay, K., Wanko, H., Ulrich, J., 2006. Absorption of Volatile Organic Compounds in Biodiesel.

Chemical Engineering Research and Design 84, 22–28. Biard, P.-F., Coudon, A., Couvert, A., Giraudet, S., 2016. A simple and timesaving method

for the mass transfer assessment of solvents used in physical absorption. Chemical Engineering Journal 290, 302–311.

Biard, P.-F., Couvert, A., Giraudet, S., 2018. Volatile organic compounds absorption in packed column: theoretical assessment of water, DEHA and PDMS 50 as absorbents. J. Indus. Eng. Chem. 59, 70–78.

Billet, R., Schultes, M., 1999. Prediction of mass transfer columns with dumped and arranged packings: updated summary of the calculation method of Billet and Schultes. Chemical Engineering Research and Design 77, 498–504.

Billet, R., Schultes, M., 1995. Fluid dynamics and mass transfer in the total capacity range of packed columns up to the flood point. Chem. Eng. Technol. 18, 371–379.

Billet, R., Schultes, M., 1993. A physical model for the prediction of liquid hold-up in two-phase countercurrent columns. Chemical engineering & technology 16, 370–375.

Billet, R., Schultes, M., 1991. Modelling of pressure drop in packed columns. Chem. Eng. Technol. 14, 89–95.

Bourgois, D., Thomas, D., Fanlo, J.-L., Vanderschuren, J., 2006. Solubilities at High Dilution of Toluene, Ethylbenzene, 1,2,4-Trimethylbenzene, and Hexane in Di-2-ethylhexyl, Diisoheptyl, and Diisononyl Phthalates. Journal of Chemical & Engineering Data 51, 1212–1215.

Bourgois, D., Vanderschuren, J., Thomas, D., 2009. Study of mass transfer of VOCs into viscous solvents in a pilot-scale cables-bundle scrubber. Chemical Engineering Journal 145, 446–452.

Brunazzi, E., Paglianti, A., Spiegel, L., Tolaini, F., 2002. Hydrodynamics of a gas-liquid column equipped with MellapakPlus packing, in: International Conference on Distillation and Absorption.

Cherif, H., Coquelet, C., Stringari, P., Clodic, D., Toubassy, J., 2017. Comparison of models for the prediction of hydrodynamic parameters in structured packing columns for biogas purification.

Darracq, G., 2011. Couplage de l’absorption dans une phase organique et de la biodégradation dans un réacteur multiphasique. Application au traitement de Composés Organiques Volatils hydrophobes. PhD thesis ENSCR-005, Rennes.

Darracq, G., Couvert, A., Couriol, C., Amrane, A., Le Cloirec, P., 2010. Integrated process for hydrophobic VOC treatment-solvent choice. The Canadian Journal of Chemical Engineering 655–660.

Darracq, Guillaume, Couvert, A., Couriol, C., Amrane, A., Thomas, D., Dumont, E., Andres, Y., Le Cloirec, P., 2010. Silicone oil: An effective absorbent for the removal of hydrophobic volatile organic compounds. Journal of Chemical Technology & Biotechnology 85, 309–313.

Guihéneuf, S., Castillo, A.S.R., Paquin, L., Biard, P.-F., Couvert, A., Amrane, A., 2014. Absorption of Hydrophobic Volatile Organic Compounds in Ionic Liquids and Their Biodegradation in Multiphase Systems, in: Fang, Z., Smith, Jr., Richard L., Qi, X. (Eds.), Production of Biofuels and Chemicals with Ionic Liquids, Biofuels and Biorefineries. Springer Netherlands, Dordrecht, pp. 305–337.

Jour

nal P

re-p

roof

18

Guillerm, M., 2017. Optimisation du couplage de l’absorption et de la biodégradation pour l’élimination de Composés Organiques Volatils hydrophobes (These de doctorat). Rennes, Ecole nationale supérieure de chimie.

Guillerm, M., Couvert, A., Amrane, A., Norrant, E., Lesage, N., Dumont, E., 2016. Absorption of toluene in silicone oil: Effect of the solvent viscosity on hydrodynamics and mass transfer. Chemical Engineering Research and Design 109, 32–40.

Hadjoudj, R., Monnier, H., Roizard, C., Lapicque, F., 2004. Absorption of Chlorinated VOCs in High-Boiling Solvents: Determination of Henry’s Law Constants and Infinite Dilution Activity Coefficients. Industrial & Engineering Chemistry Research 43, 2238–2246.

Heymes, F., Manno Demoustier, P., Charbit, F., Louis Fanlo, J., Moulin, P., 2006. Hydrodynamics and mass transfer in a packed column: Case of toluene absorption with a viscous absorbent. Chemical Engineering Science 61, 5094–5106.

Kampa, M., Castanas, E., 2008. Human health effects of air pollution. Environmental Pollution, Proceedings of the 4th International Workshop on Biomonitoring of Atmospheric Pollution (With Emphasis on Trace Elements) 151, 362–367.

Lalanne, F., Malhautier, L., Roux, J.-C., Fanlo, J.-L., 2008. Absorption of a mixture of volatile organic compounds (VOCs) in aqueous solutions of soluble cutting oil. Bioresource Technology 99, 1699–1707.

Le Cloirec, P., 2004. COV (composés organiques volatils). Techniques de l’Ingénieur G1, 1–10.

Lhuissier, M., Couvert, A., Amrane, A., Kane, A., Audic, J.-L., 2018. Characterization and selection of waste oils for the absorption and biodegradation of VOC of different hydrophobicities. Chemical Engineering Research and Design 138, 482–489.

Mackowiak, J., 2010. Fluid Dynamics of Packed Columns - Principles of the Fluid Dynamic Design of Columns for Gas/Liquid and Liquid/Liquid Systems, VDI-Buch. Springer, Berlin Heidelberg.

Maćkowiak, J., 1991. Pressure drop in irrigated packed columns. Chemical Engineering and Processing: Process Intensification 29, 93–105.

Maćkowiak, J., 1990. Determination of flooding gas velocity and liquid hold-up at flooding in packed columns for gas/liquid systems - Maćkowiak - 1990 - Chemical Engineering & Technology - Wiley Online Library.

Minne, U.L., 2017. Effect of liquid and gas physical properties on the hydrodynamics of packed columns (Thesis). Stellenbosch : Stellenbosch University.

Minne, U.L., Burger, A.J., Schwarz, C.E., 2018. The effect of fluid properties and packing size on the hydrodynamics of packed columns. Chemical Engineering Transactions 69, 31–36.

Miyahara, T., Ogawa, K., Nagano, Y., Hirade, A., Takahashi, T., 1992. Flow dynamics in low height packed column having large fractional void space. Chemical engineering science 47, 3323–3330.

Ozturk, B., Yilmaz, D., 2006. Absorptive Removal of Volatile Organic Compounds from Flue Gas Streams. Process Safety and Environmental Protection 84, 391–398.

Piché, S., Larachi, F., Grandjean, B., 2001a. Improving the prediction of irrigated pressure drop in packed absorption towers. Can. J. Chem. Eng. 79, 584–594.

Piché, S., Larachi, F., Grandjean, B., 2001b. Loading Capacity in Packing Towers–Database, Correlations and Analysis. Chem. Eng. Technol. 24, 373–380.

Piché, S., Larachi, F., Grandjean, B., 2001c. Flooding capacity in packed towers: database, correlations, and analysis. Ind. Eng. Chem. Res. 40, 476–487.

Rocha, J.A., Bravo, J.L., Fair, J.R., 1993. Distillation columns containing structured packings: a comprehensive model for their performance. 1. Hydraulic models. Industrial & Engineering Chemistry Research 32, 641–651.

Rodriguez Castillo, A.S., Biard, P.-F., Guihéneuf, S., Paquin, L., Amrane, A., Couvert, A., 2019. Assessment of VOC absorption in hydrophobic ionic liquids: Measurement of partition and diffusion coefficients and simulation of a packed column. Chem. Eng. J. 360, 1416–1426.

Jour

nal P

re-p

roof

19

Rodriguez Castillo, A.S., Guihéneuf, S., Biard, P.-F., Paquin, L., Amrane, A., Couvert, A., 2018. Physicochemical properties of some hydrophobic room-temperature ionic liquids applied to volatile organic compounds biodegradation processes: Physicochemical properties of some hydrophobic room-temperature ionic liquids. Journal of Chemical Technology & Biotechnology 93, 215–223.

Roustan, M., 2003. Transferts gaz-liquide dans les procédés de traitement des eaux et des effluents gazeux. Lavoisier, Paris.

Ruddy, E.N., Carroll, L.A., 1993. Select the best VOC control strategy. Chemical engineering progress 89, 28–35.

Stichlmair, J., Bravo, J.L., Fair, J.R., 1989. General model for prediction of pressure drop and capacity of countercurrent gas/liquid packed columns. Gas Separation & Purification 3, 19–28.

Tsai, R.E., Seibert, A.F., Eldridge, R.B., Rochelle, G.T., 2009. Influence of viscosity and surface tension on the effective mass transfer area of structured packing. Energy Procedia 1, 1197–1204.

Zakeri, A., Einbu, A., Svendsen, H.F., 2012. Experimental investigation of liquid holdup in structured packings. Chemical Engineering Research and Design 90, 585–590.

Jour

nal P

re-p

roof

20

Figure 1: Experimental set-up.

Figure 2 : Evolution of the pressure drop with the gas velocity for PDMS (a), transformer oil (b) and

lubricant (c) for increasing values of the liquid load UL (in m3 m-2 h-1). The straight lines correspond to the

calculations according to the Billet-Schultes correlations (part 3.3).

Air

Fan

T P F

DP

Centrifugal pump

Solvant tank

U-tube filledwith water

Heat exchanger

Packed column(120 mm ID, packing height = 1 m

Flexipac® packing)

Air outlet

T

F

0

50

100

150

200

250

300

350

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

DP/

Z (P

a m

-1)

UG (m s-1)

1.4

2.7

4.0

5.2

6.5

7.7

8.9

Loading

0

50

100

150

200

250

300

350

400

450

0.0 0.2 0.4 0.6 0.8 1.0 1.2

DP/

Z (P

a m

-1)

UG (m s-1)

4.3

9.9

16.1

22.79

29.8

37.1

Loading

Flooding

0

50

100

150

200

250

300

350

400

0.0 0.2 0.4 0.6 0.8 1.0 1.2

DP/

Z (P

a m

-1)

UG (m s-1)

3.1

7.4

12.4

17.8

23.7

29.8

36.3

Loading

Flooding

(a) PDMS 20 (b) Transformer oil (c) Lubricant

UL (m h-1) UL (m h-1) UL (m h-1)

Jour

nal P

re-p

roof

21

Figure 3: Evolution of the loading and flooding gas velocity for transformer oil (a) and lubricant (b) for

increasing values of the L/G ratio (Flexipac® 500Z HC). The green frame corresponds to the loading zone

(located between the loading and the flooding velocities). The orange and blue lines correspond to 80%

and 60% of the flooding velocity, respectively. These data were simulated using the B-S correlations.

Figure 4: Evolution of the pressure drop (orange curves), the liquid holdup (yellow curves) and the ratio

a°/a (green curves) for increasing values of the gas velocity for transformer oil (a) and lubricant (b)

(Flexipac® 500Z HC). Two liquid loads (5 and 20 m h-1) were simulated for the transformer oil and one

liquid load (5 m h-1) was simulated for the lubricant. These data were simulated using the B-S

correlations.

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

1.3

1.4

0.5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

UG

(m s

-1)

L/G (kg kg-1)

Lubricant flooding

0,8fl

0,6ufl

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

1.3

1.4

0.5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

UG

(m s

-1)

L/G (kg kg-1)

Transformer oil flooding

0,8fl

0,6ufl

(a) (b)

Loading zone (transformer oil)0.8×UG,Fl

0.6×UG,Fl

Loading zone (lubricant)0.8×UG,Fl

0.6×UG,Fl

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

110%

120%

0

50

100

150

200

250

300

350

400

450

500

550

600

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1

hL

and

a0 /

a

DP/

Z (P

a m

-1)

UG (m s-1)

DP/Z

hL

a°/a

0%

15%

30%

45%

60%

75%

90%

105%

120%

135%

150%

0

50

100

150

200

250

300

350

400

450

500

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2

hL

and

a0 /

a

DP/

Z (P

a m

-1)

UG (m s-1)

DP/Z

DP/Z

hL

a°/a

hL

a°/a

(a) (b)

DP/ZhLa°/a

DP/ZhLa°/a

DP/ZhLa°/a

UL = 5 m h-1

UL = 5 m h-1

UL = 20 m h-1

Jour

nal P

re-p

roof

22

Figure 5: Evolution of the pressure drop (black curve), the liquid holdup (orange curve), the interfacial

area a° (purple curve), the column diameter (green curve) and the liquid (blue curve) and gas (red curve)

velocities for increasing values of the L/G ratio for the transformer oil for the following design: FG = 4000

Nm3 h-1, UG = 0.80×UG,Fl, using Flexipac® 500Z HC and G = 1.17 kg m-3.

0

40

80

120

160

200

240

280

320

360

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

1 2 3 4 5

DP/

Z (P

a m

-1);

uL

(m h

-1) a

nd

a°

(m2 m

-3)

UG

(m s

-1);

Dco

l(m

) an

d h

L

L/G (kg kg-1)

UG

Dcol

hL

UL

a°

DP

Dcol

a°

UG

DP/Z

hL

UL

Jour

nal P

re-p

roof

23

Table 1: Properties of the studied liquid absorbents.

PDMS Transformer oil Lubricant

Chemical composition Siloxanes Linear alkanes

Dynamic viscosity L (mPa s) at 25°C 20 19 79

Molar mass (g mol-1) 3000 212*

Density L (kg m-3) 900 865 875

Superficial tensionL (N m-1) 0.021 0.027 0.031

* Lubricant and transformer oil have complex chemical structures. Several samples were investigated by 1H-NMR spectroscopy and showed quite similar chemical natures between lubricant and transformer oil and allowed to roughly estimate an average chain length of 15 carbons which is coherent with literature data for mineral oils (Aluyor and Ori-jesu, 2009).

Table 2: Packing characteristics

Name dh (m) a (m−1) ε (-)

Flexipac® 500Z HC 7.60.10-3 500 0.95

Table 3: Operating conditions

P T D Z FG FL UG UL

bar °C m m Nm3 h-1 L h-1 m s-1 m s-1

1 25 0.12 1 9 – 50 15 – 750 0.2 – 1.2 3.10-4 – 2.10-2

Jour

nal P

re-p

roof

24

Table 4: Billet-Schultes correlations developed for the determination of the parameters relative to the

hydrodynamics of an irrigated counter-current packed column (Billet and Schultes, 1999, 1995, 1993,

1991).

Determination of the loading point (UG,Lo) for a given L/G ratio

𝑈𝐺,𝐿𝑜 = √𝑔

𝜓𝐿𝑜× (𝜀 − ℎ𝐿,𝐿𝑜) × √

ℎ𝐿,𝐿𝑜

𝑎× √

𝜌𝐿

𝜌𝐺

𝑈𝐿,𝐿𝑜 =𝜌𝐺

𝜌𝐿

×𝐿

𝐺× 𝑈𝐺,𝐿𝑜

With:

𝜓𝐿𝑜 =𝑔

𝐶𝐿𝑜2 × (

𝐿

𝐺.√

𝜌𝐺

𝜌𝐿(

𝜇𝐿

𝜇𝐺)

0.4

)0,752

if 𝐿

𝐺× √

𝜌𝐺

𝜌𝐿≤ 0.4 (liquid dispersed in the gas)

𝜓𝐿𝑜 =𝑔

(0.695𝐶𝐿𝑜(𝜇𝐿𝜇𝐺

)0.1588

)

2 × (𝐿

𝐺√

𝜌𝐺

𝜌𝐿(

𝜇𝐿

𝜇𝐺)

0.4

)1,46

if 𝐿

𝐺× √

𝜌𝐺

𝜌𝐿> 0.4 (gas dispersed in the

liquid) And:

ℎ𝐿,𝐿𝑜 = (12

𝑔

𝜇𝐿

𝜌𝐿

𝑎2𝑈𝐿,𝐿𝑜)1 3⁄

× (𝑎ℎ

𝑎)

2 3⁄

With:

(𝑎ℎ

𝑎) = 𝐶ℎ × (

𝑈𝐿,𝐿𝑜.𝜌𝐿

𝑎.µ𝐿)

0.15

× (𝑈𝐿,𝐿𝑜

2.𝑎

𝑔)

0.1

if (𝑈𝐿,𝐿𝑜.𝜌𝐿

𝑎.µ𝐿) < 5

(𝑎ℎ

𝑎) = 0.85 × 𝐶ℎ × (

𝑈𝐿,𝐿𝑜.𝜌𝐿

𝑎.µ𝐿)

0.25

× (𝑈𝐿,𝐿𝑜

2.𝑎

𝑔)

0.1

if (𝑈𝐿,𝐿𝑜.𝜌𝐿

𝑎.µ𝐿) ≥ 5

Eq. 1 Eq. 2 Eq. 3 Eq. 4 Eq. 5 Eq. 6 Eq. 7

Determination of the flooding point (UG,Fl) for a given L/G ratio

ℎ𝐿,𝐹𝑙3 . (3. ℎ𝐿,𝐹𝑙 − 𝜀) =

6

𝑔. 𝑎2. 𝜀.

𝜇𝐿

𝜌𝐿

.𝐿

𝐺.𝜌𝐺

𝜌𝐿

. 𝑈𝐺,𝐹𝑙

With:

𝑈𝐺,𝐹𝑙 = √2. 𝑔

𝜓𝐹𝑙

×(𝜀 − ℎ𝐿,𝐹𝑙)

1.5

𝜀0.5× √

ℎ𝐿,𝐹𝑙

𝑎× √

𝜌𝐿

𝜌𝐺

And:

𝜓𝐹𝑙 =𝑔

𝐶𝐹𝑙2 . (

𝐿

𝐺√

𝜌𝐺

𝜌𝐿(

𝜇𝐿

𝜇𝐺)

0.2

)0,388

if 𝐿

𝐺× √

𝜌𝐺

𝜌𝐿≤ 0.4

𝜓𝐹𝑙 =𝑔

(0.6244𝐶𝐹𝑙(𝜇𝐿𝜇𝐺

)0.1028

)

2 × (𝐿

𝐺√

𝜌𝐺

𝜌𝐿(

𝜇𝐿

𝜇𝐺)

0.2

)1.416

if 𝐿

𝐺× √

𝜌𝐺

𝜌𝐿> 0.4

Eq. 8 Eq. 9 Eq. 10 Eq. 11

Determination of the holdup (hL) at the working point (UG < UG,Fl and

𝑼𝑳 =𝝆𝑮

𝝆𝑳.

𝑳

𝑮. 𝑼𝑮)

Step 1: hL,Lo from Eqs. 5-7 and calculated with UL

Step 2: ℎ𝐿,𝐹𝑙 = 2.2ℎ𝐿,𝐿𝑜 (𝜇𝐿𝜌𝑊

𝜇𝑊𝜌𝐿)

0.05

Step 3: ℎ𝐿 = ℎ𝐿,𝐿𝑜 + (ℎ𝐿,𝐹𝑙 − ℎ𝐿,𝐿𝑜) × (𝑈𝐺

𝑈𝐺,𝐹𝑙)

13

(with hL,Lo and hL,Fl from steps 1 and 2)

Eq. 12 Eq. 13 Eq. 14

Determination of the pressure drop (DP/Z) at the working point

𝛥𝑃

𝑍= 𝜓𝐿 ×

𝑎

(𝜀 − ℎ𝐿)3×

𝑈𝐺2𝜌𝐺

2× (1 +

4

𝑎𝐷𝑐𝑜𝑙

)

With:

𝜓𝐿 = 𝐶𝑃 × (64

𝑅𝑒𝐺+

1.8

𝑅𝑒𝐺0.08) × 𝑒𝑥𝑝 (

13300

𝑎. √

𝑈𝐿2

𝑔) × (

𝜀−ℎ𝐿

𝜀)

1.5

× (ℎ𝐿

ℎ𝐿,𝐿𝑜)

0.3

with hL and hL,Lo

calculated from Eqs. 12 and 14

𝑅𝑒𝐺 =6

𝑎

𝑈𝐺𝜌𝐺

𝜇𝐺

(1 +4

𝑎𝐷𝑐𝑜𝑙

)−1

Eq. 15 Eq. 16 Eq. 17

Determination of the interfacial area (a°) at the working point

Jour

nal P

re-p

roof

25

Step 1: (𝑎°

𝑎)

𝐿𝑜= 1.5(𝑎𝑑ℎ)−0.5 (

𝑈𝐿𝑑ℎ𝜌𝐿

µ𝐿)

−0.2

(𝑈𝐿

2𝑑ℎ𝜌𝐿

σ𝐿)

0.75

(𝑈𝐿

2

𝑔𝑑ℎ)

0.45

in which UL is the

working liquid velocity

Step 2: (𝑎°

𝑎)

𝐹𝑙= 7 (

𝑎°

𝑎)

𝐿𝑜(

σ𝐿

σ𝑊)

0.56

Step 3: (𝑎°

𝑎) = (

𝑎°

𝑎)

𝐿𝑜+ ((

𝑎°

𝑎)

𝐹𝑙− (

𝑎°

𝑎)

𝐿𝑜) × (

𝑈𝐺

𝑈𝐺,𝐹𝑙)

13

(with (𝑎°

𝑎)

𝐿𝑜and (

𝑎°

𝑎)

𝐹𝑙from steps 1

and 2)

Eq. 18 Eq. 19 Eq. 20

Table 5. Comparison of the experimental loading gas velocities (UG,Lo) at a given liquid load (UL) and L/G

ratio to the ones predicted using the Billet-Schultes correlations (Eqs 1-7). RE corresponds to the relative

error between the experimental and the theoretical loading gas superficial velocities. The liquid holdup

(hL,Lo) determined with the Billet-Schultes correlations is also provided.

UL (m h-1) L/G UG,Lo experimental (m s-1)

UG,Lo model (m s-1)

RE hL,Lo model

PDMS

3.06 1.02 0.64 0.64 0.0% 0.092

7.38 2.81 0.56 0.55 2.4% 0.150

12.36 5.50 0.48 0.48 0.5% 0.203

23.65 12.95 0.39 0.36 8.2% 0.280

Transformer oil 9.92 3.18 0.64 0.53 17.8% 0.160

16.14 5.92 0.56 0.46 17.6% 0.211

Lubricant

1.40 0.45 0.65 0.66 1.2% 0.083

2.69 1.00 0.56 0.59 5.0% 0.122

3.97 1.72 0.48 0.54 11.6% 0.158

5.23 2.53 0.43 0.50 15.3% 0.188

Jour

nal P

re-p

roof

26

Table 6. Comparison of the experimental flooding gas velocities (UG,Fl) at given liquid load (UL) and L/G

ratio to the ones predicted using the Billet-Schultes correlations (Eqs 8-11). RE corresponds to the

relative error between the experimental and the theoretical flooding gas superficial velocities. The liquid

holdup (hL,Fl) determined with the Billet-Schultes correlations is also provided.

UL (m h-1) L/G UG,Fl experimental (m s-1)

UG,Fl model (m s-1)

RE hL,Fl model

PDMS

17.81 3.59 1.06 0.92 13.4% 0.3927

23.65 6.24 0.81 0.80 1.5% 0.4145

29.82 8.72 0.73 0.73 0.1% 0.4293

36.27 13.83 0.56 0.58 3.5% 0.4445

Transformer oil

16.14 3.38 0.98 0.91 7.1% 0.3909

22.79 5.78 0.81 0.80 1.7% 0.4116

29.78 9.40 0.65 0.70 7.5% 0.4331

Lubricant

6.48 1.26 1.07 1.00 5.7% 0.4181

7.71 1.95 0.82 0.90 9.2% 0.4382

8.94 2.54 0.73 0.83 13.7% 0.4511

10.16 3.25 0.65 0.77 19.0% 0.4637

Table 7: Billet-Schultes constants determined and comparison with the ranges determined by the authors

for other packing materials.

a (m-

1) ε CP CLo Ch CFl

Flexipac® 500Z HC 500 0.95 0.462 1.68 1.50 1.73

Range of B-S correlations (Billet and Schultes, 1999)

80-545

0.68-0.985

0.25-1.33

2.45-3.79

0.482-1.90

1.55-3.02

Table 8: ARE between the experimental and theoretical (B-S correlations) pressure drops for the three

solvents studied in the pre-loading and loading zones.

PDMS 20 Transformer oil Lubricant

Both zones

Pre-loading zone

Loading zone

Both zones

Pre-loading zone

Loading zone

Both zones

Pre-loading zone

Loading zone

17.9% 31.8% 12.1% 19% 32% 13% 27% 36% 20%

Jour

nal P

re-p

roof